Gemma 4 31B Architecture
Back to Gemma 4 31B in the gallery
Dense Gemma 4 scales the family to a 256K-context multimodal checkpoint without changing the core local-global recipe much.
Architecture and specifications
- Scale
- 30.7B parameters
- Context (tokens)
- 262,144
- Vocabulary
- 262,144 (~262k)
- License
- Apache License 2.0
- Date
- 2026-04-02
- Decoder type
- Dense
- Attention
- GQA with QK-Norm, unified K/V on global layers, p-RoPE on global layers, and 5:1 sliding-window/global attention
- Layer mix
- 50 sliding-window + 10 global
- KV cache / token (bf16) info
- 850 KiB (compact global K=V layout) · Very high
- Key detail
- Carries Gemma's unusual pre/post-norm stack into a larger 31B dense model with 256K context.
Training and benchmarks
- Pretraining data
- Not disclosed No per-model pretraining-token count is published. Gemma 4 model card
- Post-training methods
- Instruction tuning IT variant; the report refers to Gemma 3-like post-training without itemizing alignment stages. Gemma 4 report